Logit Modeling and Logistic Regression: Aphids, Ants, and Plants

Abstract Ecological data are often discontinuous or categorical. This means that measurement variables tend to have values like “present vs. absent” or “treatment vs. control.” The example that I consider in detail in this chapter involves categorical data at nearly every level of analysis: patterns of association among predatory ants in the genus Pheidole, herbivorous aphids in the genus Aphis, and host plants in the genus Piper. Categorical measurements include experimental manipulation of ant abundance (ants present vs. ants removed), experimental manipulation of host plant quality (fertilized plants vs. control plants), observations of aphid abundance (colony present vs. colony absent), characterization of aphid diet breadth (specialist vs. generalist), year of study (year 1 vs. year 2), and so forth.